itmo-dev: Elderly People Safety App 🌟

Welcome to the itmo-dev repository, home of the Elderly People Safety App. This project aims to enhance the safety of elderly individuals through a comprehensive application leveraging machine learning models and various cutting-edge technologies.

Project Features 🚀

• 🔒 SSL Certificates: The project ensures secure communication through SSL certificates.
• 🛡️ DDoS Protection: Implemented protection against DDoS attacks using rate limiting.
• ⚙️ Asynchronous Queues: Utilizes asynchronous queues with the ARQ library for efficient task handling.
• 📊 Data Visualization: Employs Streamlit for data visualization, making insights accessible and understandable.
• 🌐 Microservices Architecture: The project follows a microservices architecture for scalability and maintainability.
• 💳 Billing: deduction of credits for forecasts, different cost of models, 100 free credits for registration

Demo and Documentation 📚

• Demo: Check out the live demo here.
• API Documentation: Explore the API documentation here.

Machine Learning Models 🤖

The following machine learning models were developed:

1. LightGBM Model: Metric - 0.79
2. Decision Tree Model: Metric - 0.82
3. Logistic Regression Model: Metric - 0.836

Detailed results of the research are available in the research_result folder.

Getting Started 🛠️

Prerequisites

• Docker

Running the Project ▶️

1. Clone the repository:

   git clone https://github.com/KaroUniform/itmo-dev
   cd itmo-dev

2. Build and run the project using Docker Compose:

   sudo docker-compose up
   

Scaling 📈

You can scale the project by adjusting the number of workers and API instances:

Scale worker (adjust the value of N):

sudo docker-compose up --scale worker=N

Scale API in gunicorn_conf.py (adjust the value of N):

workers = N

Or ⚠️(Not recommended, not tested) :

sudo docker-compose up --scale api=N

🌐 Database schema

users

Column	Type	Constraints
id	UUID	Primary Key
balance	Float	Default: 0

transaction_history

Column	Type	Constraints
job_id	UUID	Primary Key, Index
user_id	UUID	Foreign Key (users.id)
amount	Integer	Not Null
model_id	Integer	Foreign Key (ml_models.id), Nullable
result	JSON	Nullable
status	Enum	Default: IN_PROGRESS
err_reason	String(512)	Nullable
timestamp	DateTime	Server Default: func.now()

ml_models

Column	Type	Constraints
id	Integer	Primary Key, Index
model_name	String	Unique
model_cost	Float

Triggers

1. After Insert on transaction_history Table:

• This trigger is executed after a new record is inserted into the transaction_history table.
• It asynchronously fetches the associated user and updates their balance by adding the transaction amount.

2. Before Update on transaction_history Table:

• This trigger is executed before an update operation on the transaction_history table.
• If the transaction status is marked as FAILURE, it asynchronously retrieves the associated user.
• It checks for the change in status to FAILURE and, if detected, refunds the user by adding the absolute amount of the transaction to their balance.
• Additionally, it updates the transaction amount to zero in the transaction_history table.

3. After Insert on users Table:

• This trigger is executed after a new user is inserted into the users table.
• It creates a new transaction record with a predefined amount, status, and result for the user.
• The new transaction is then added to the database asynchronously.